Two tools for two questions
Google Analytics 4 mainly answers questions about usage: which pages are read, which channels bring visitors, at which step of a funnel they drop off? GA4 records interactions as events and offers explorations for funnels, paths and cohorts next to the standard reports.
TSO answers the question that comes after the form: which campaign brings leads that turn into booked calls, closed deals and revenue? To do that, TSO connects the click on the ad with events that happen outside the website, such as a deal won in the CRM or a payment in Stripe. These results go back to the ad platforms through the conversion API, so Meta, Google Ads and others optimize for closed deals instead of form fills.
See which ad brings customers, not just leads.
Leads, deals and revenue per channel, campaign and ad. Seven attribution models, any window, switchable at any time.
Using GA4 and TSO together
You don't have to remove GA4 to use TSO. The two scripts run independently.
- Install the script: Add the TSO script directly or through Google Tag Manager, next to the Google tag. Events from the dataLayer can be taken over in TSO with a rule, without code.
- Align consent: TSO detects Google Consent Mode v2 and banners such as Cookiebot, Usercentrics or Borlabs. You choose one of three privacy modes per project.
- Connect your deals: Send booked calls and won deals by webhook, API or CSV import, and take over payments from Stripe, Digistore24 or CopeCart. TSO matches them to the lead by email or phone.
- Sort out Google Ads: If you import key events from GA4 into Google Ads and also send the same action through TSO, the conversion arrives twice. Decide in Google Ads which conversion action counts for bidding (primary) and keep the other for observation only (secondary).
- Connect the other platforms: Meta, TikTok, LinkedIn, Microsoft Ads, Snapchat and ChatGPT Ads get their conversions through the TSO conversion API, with the same event ID as the pixel.
Why GA4 and TSO show different numbers
Differences are normal, because each tool counts by its own rules. The most common reasons:
| Reason | Google Analytics 4 according to Google | TSO |
|---|---|---|
| Attribution model | Data-driven or one of the two last click models | The dashboard's model, for example last paid click |
| Attribution window | By default 30 days for acquisition key events and 90 days for all others | The dashboard's window, from 1 day to unlimited |
| Visit or session | Times out after 30 minutes of inactivity by default | Ends after 45 minutes of inactivity or with a new campaign source |
| Direct visits | Only receive credit if the path consists entirely of direct visits | Depends on the model. Unattributed conversions are shown as a row of their own |
| Consent | With consent mode, GA4 can model the behavior of users without analytics cookies once the minimums Google names are met | Depends on the privacy mode, in the restricted mode nothing is recorded without consent |
| What counts as a conversion | Key events on website and app, offline events through data import or the Measurement Protocol | Leads and system events such as booked call, deal won or payment from CRMs and payment providers. Duplicates (reload, second tab, same order ID) are suppressed |
If you want to reconcile the two directly, set TSO up as close to GA4 as possible: last non-direct click comes closest to "paid and organic last click", with the same window. For budget decisions, what counts in the end is which campaign brings closed deals, and only the data from sales shows that.
How each model splits a conversion is explained in the guide to attribution models.
Sources
What this page says about Google Analytics 4 comes from Google itself (Google Marketing Platform, Google Analytics Help and developer documentation):
- Google Analytics, product page
- Introducing Google Analytics 4
- Setup for websites and apps
- Key events and recommended events
- User-ID and reporting identity
- Attribution models and attribution settings
- Sessions
- Linking Google Ads and other integrations
- Data import
- Behavioral modeling for consent mode
- Explorations
- BigQuery export
- Measurement Protocol
- Data API and MCP server